Biochemistry 1997, 36, 7917-7927
7917
Rapid and Parallel Formation of Fe3+ Multimers, Including a Trimer, during H-Type Subunit Ferritin Mineralization† Alice S. Pereira,‡ Pedro Tavares,‡ Steven G. Lloyd,‡ Dana Danger,§ Dale E. Edmondson,*,| Elizabeth C. Theil,*,§ and Boi Hanh Huynh*,‡ Departments of Physics, Biochemistry, and Chemistry, Rollins Research Building, Emory UniVersity, Atlanta, Georgia 30322, and Department of Biochemistry, North Carolina State UniVersity, Raleigh, North Carolina 27695 ReceiVed February 14, 1997X
ABSTRACT:
Conversion of Fe ions in solution to the solid phase in ferritin concentrates iron required for cell function. The rate of the Fe phase transition in ferritin is tissue specific and reflects the differential expression of two classes of ferritin subunits (H and L). Early stages of mineralization were probed by rapid freeze-quench Mo¨ssbauer, at strong fields (up to 8 T), and EPR spectroscopy in an H-type subunit, recombinant frog ferritin; small numbers of Fe (36 moles/mol of protein) were used to increase Fe3+ in mineral precursor forms. At 25 ms, four Fe3+-oxy species (three Fe dimers and one Fe trimer) were identified. These Fe3+-oxy species were found to form at similar rates and decay subsequently to a distinctive superparamagentic species designated the “young core.” The rate of oxidation of Fe2+ (1026 s-1) corresponded well to the formation constant for the Fe3+- tyrosinate complex (920 s-1) observed previously [Waldo, G. S., & Theil, E. C. (1993) Biochemistry 32, 13261] and, coupled with EPR data, indicates that several or possibly all of the Fe3+-oxy species involve tyrosine. The results, combined with previous Mo¨ssbauer studies of Y30F human H-type ferritin which showed decreases in several Fe3+ intermediates and stabilization of Fe2+ [Bauminger, E. R., et al. (1993) Biochem. J. 296, 709], emphasize the involvement of tyrosyl residues in the mineralization of H-type ferritins. The subsequent decay of these multiple Fe3+-oxy species to the superparamagnetic mineral suggests that Fe3+ species in different environments may be translocated as intact units from the protein shell into the ferritin cavity where the conversion to a solid mineral occurs.
Ferritin is the only protein known to direct a reversible phase transition from a metal ion in solution to a metal ion in a solid matrix (Waldo & Theil, 1996). Bone and tooth formation and resorption processes are formally analogous to the mineralization of ferritin, but a large number of cells and extracellular macromolecules are required for mineralization, in contrast to the case for ferritin. The major role of ferritin in all cells is to concentrate and store iron, since the metal is used in amounts that are effectively ∼1011 times the solubility of the free ion. During oxidative stress or with an iron excess, ferritin appears to protect against radical damage by managing iron-dioxygen interactions (Touati et al., 1995; Balla et al., 1992). The structure of ferritin is common to all known organisms (animals, plants, and microorganisms): a hollow sphere with a diameter of ∼12 nm formed by 24 protein subunits, each of which is a fourhelix bundle (Banyard et al., 1978; Harrison & Lilley, 1990; Trihka et al., 1994). The mineral grows and dissolves within the hollow (diameter of ∼8 nm) at the center of the protein (Waldo & Theil, 1996; Harrison & Arosio, 1996). † This work was supported in part by National Institutes of Health grants GM 47295 (to B.H.H. and D.E.E.) and DK 20251 (to E.C.T.). S.G.L. is a recipient of a Young Scientist Training Scholarship from the Life and Health Insurance Medical Research Fund. A.S.P. and P.T. are recipients of postdoctoral research fellowships from Programa PRAXIS XXI of Junta Nacional de Investigacao Cientifica e Tecnologica, Portugal. * Authors to whom correspondence should be addressed. ‡ Department of Physics, Emory University. § North Carolina State University. | Departments of Biochemistry and Chemistry, Emory University. X Abstract published in AdVance ACS Abstracts, June 15, 1997.
S0006-2960(97)00348-6 CCC: $14.00
Rates of ferritin mineralization depend on the ratio of Hand L-type ferritin subunits present in ferritin, which is varied in animals by the differential expression of the two sets of ferritin genes, H and L, in various tissues (Theil, 1987; Harrison & Arosio, 1996; Waldo & Theil, 1996). Studies with recombinant proteins composed of solely H or L subunits have shown that H-type subunits have very fast rates of iron uptake (Fe oxidation and hydrolysis) (Treffry et al., 1993; Waldo et al., 1993) compared to those of L-type subunit proteins; rates differ by more than 100-fold (Waldo & Theil, 1993). Initial rates of mineralization in ferritin from bacteria are also rapid (LeBrun et al., 1993), although slower than the H-subunit ferritin from animals (Waldo & Theil, 1996). The rapid rate of ferritin Fe uptake is accompanied by the transient formation of a Fe3+-tyrosinate complex (Waldo et al., 1993). In general, H-type ferritin subunits predominate in the ferritin of tissues with high oxygen levels (e.g., cardiac and erythroid cells) while L-type ferritin subunits predominate in the ferritin of tissues with slower iron turnover (e.g., liver). Intermediates in the conversion of hexaquo Fe2+ to solid hydrated ferric oxide in ferritin include multimers of Fe3+oxo complexes bound to the protein, detected by EXAFS and Mo¨ssbauer spectroscopy (Yang et al., 1987; Bauminger et al., 1989, 1991, 1993). These earlier studies, however, investigated the properties of Fe3+ intermediates at times now known to be long in comparison with the formation time of the initial Fe-protein complex in ferritins composed of H-type subunits. In order to explore the ferritin Fe uptake mechanism and to characterize the type of Fe3+ intermediates that form after ferroxidation and during the initial mineral© 1997 American Chemical Society
7918 Biochemistry, Vol. 36, No. 25, 1997
Pereira et al.
ization process, recombinant frog H-type subunit ferritin samples were rapidly mixed with oxygenated solutions of Fe2+ and quenched by rapid freezing at various times after mixing for Mo¨ssbauer spectroscopic investigations. The results show that several multinuclear (n ) 2-3) Fe3+-oxy clusters form in parallel with the rapid oxidation of Fe2+. The rate of the rapid Fe2+ oxidation by H-subunit ferritin was determined directly for the first time and was found to be similar to that of the formation of the Fe-tyrosine complex determined earlier using stopped-flow techniques (Waldo & Theil, 1993). These observations demonstrate that multiple processes are involved immediately after the initial oxidation of Fe2+ by ferritin and that the purple Fe-tyrosine complex (or complexes) identified previously is an initial product of the oxidation. Further, data of a ferritin sample reacted with Fe2+ for 2 days indicate clearly that these initially forming complexes are precursors of the ferritin mineral. Implications of our current findings in relation to ferritin function are discussed. METHODS Protein Preparation. H-type subunit frog ferritin was isolated by protein expression using a pET3 vector containing the 5F12 coding sequence (Didsbury et al., 1986) in the NdeI site; host Escherichia coli BL21-3 cells were grown in medium containing sorbital and betaine (Blackwell & Horgan, 1991). Ferritin synthesis was induced in mid-log cells using IPTG (isothiopropyl thiogalactoside), and the soluble ferritin was isolated, after sonication of the bacterial cells, by FPLC on a Fast Q-Sepharose (Pharmacia) matrix. Protein purified to homogeneity was sterilized by filtration through a 0.45 µm membrane and stored at 4 °C for less than 1 week before mixing with Fe2+. Preparation of Rapid Freeze-Quenched Samples. Rapid freeze-quenched Mo¨ssbauer and EPR samples were obtained using an Update Instruments rapid freeze-quench apparatus following procedures described previously (Ravi et al., 1994). Apoferritin solutions, in 0.2 M MOPS buffer (pH 7) and 0.2 M NaCl, were oxygenated to saturation under 1 atm of O2 and then rapidly mixed with O2-saturated FeSO4 solutions, in 0.2 M NaCl, at 23 °C. The protein was allowed to react with the Fe and oxygen for fixed time periods (25 ms to 1 s) before being sprayed into cold isopentane at -140 °C. The frozen samples were packed into either EPR quartz tubes or a Mo¨ssbauer delrin cuvette and stored in liquid nitrogen. Mo¨ssbauer samples were made with an 57Fe-enriched (g95% plus enrichment) FeSO4 solution that was prepared from an iron metal foil. The final concentration of Fe was 2 mM for the Mo¨ssbauer samples and 2.4 mM for the EPR samples. The Fe/protein molar ratio was kept at 36 in order to maximize the accumulation of the Fe-protein intermediate observed in stopped-flow measurements (Waldo & Theil, 1993). Throughout this paper, the ferritin protein concentrations are expressed as 24-subunit oligomers. Under the reaction conditions of 298 K and 1 atm, O2 has a solubility of 1.23 mM in water (Dean, 1992). Since the initial concentration of Fe2+ after mixing was 2 mM and the Fe/O2 stoichiometry was determined to be 2.0 for 36 Fe/ferritin (Waldo & Theil, 1993), there is sufficient O2 to oxidize the iron. Mo¨ ssbauer and EPR Spectroscopies. Mo¨ssbauer spectra were recorded in either a weak-field spectrometer equipped
FIGURE 1: Mo¨ssbauer spectra of rapid freeze-quenched samples from the reaction of recombinant frog H-type subunit apoferritin with 57Fe2+ (36 Fe per 24 apoferritin subunits) in the presence of O2. The data were recorded at 4.2 K with a 50 mT magnetic field applied parallel to the γ-beam. The reaction was quenched at (A) 25 ms, (B) 60 ms, (C) 130 ms, (D) 220 ms, (E) 440 ms, and (F) 1 s. The solid lines are theoretical simulations of the nonreacted Fe2+ species (see the text) showing its decay in intensity as a function of reaction time. The bracket indicates the position of the central quadrupole doublet representing the initial forming ferric products of the reaction.
with a Janis 8DT variable-temperature cryostat or a strongfield spectrometer furnished with a Janis CNDT/SC SuperVaritemp cryostat encasing an 8 T superconducting magnet. Both spectrometers operate in a constant acceleration mode in a transmission geometry. The zero velocity of the spectra refers to the centroid of a room-temperature spectrum of a metallic iron foil. The Mo¨ssbauer spectra were analyzed using the WMOSS program (WEB Research Co., Edina, MN) based on a spin Hamiltonian formalism conventionally used for Mo¨ssbauer analysis (Huynh, 1994). EPR data acquisition was performed on a Bruker ER 200D-SRC spectrometer equipped with an Oxford Instrument ESR 910 continuous-flow cryostat. The spectra were recorded at 8 K with a microwave power of 2 mW, a microwave frequency of 9.43 GHz, a modulation frequency of 100 kHz, a modulation amplitude of 1 mT, and a receiver gain of 5 × 104. RESULTS Reaction of Apoferritin with Fe2+ and O2. Figure 1 shows the time-dependent Mo¨ssbauer spectra of the reaction of apoferritin (56 µM) with ferrous ions (36 Fe/ferritin molecule) in the presence of O2. The reaction was freezequenched at 0.025, 0.060, 0.13, 0.22, 0.44, or 1 s. The spectra were recorded at 4.2 K with a 50 mT field applied parallel to the γ-beam. The progress of the reaction is clearly visible from these spectra. At the first time point (0.025 s, Figure 1A), the majority of the iron atoms (55%) are in the
Characterization of Ferroxidase Products in H-Ferritin
Biochemistry, Vol. 36, No. 25, 1997 7919
unreacted ferrous state, which exhibits a broad quadrupole doublet (solid line in Figure 1A) with parameters (quadrupole splitting ∆EQ ) 3.00 mm/s and isomer shift δ ) 1.36 mm/ s)1 typical of high-spin Fe2+. As the reaction time increases, the intensity of this ferrous doublet decreases and, after 1 s of reaction time, is almost unobservable (Figure 1F). Accompanying the decay of the Fe2+ species, a broad central doublet (marked by a bracket) increases in intensity. This central doublet is comprised of at least three partially resolved quadrupole doublets with parameters (∆EQ ≈ 0.7-1.7 mm/s and δ ≈ 0.5 mm/s) indicative of high-spin Fe3+ (see below). In addition, a broad and undefined spectrum extending from -9 to 9 mm/s is observed to gain intensity with increasing reaction time. These time-dependent spectra therefore demonstrate directly the conversion of Fe2+ ions to Fe3+ species by ferritin using O2. In the results which follow, the kinetics of oxidation of Fe2+ will be presented first, followed by descriptions of various Fe3+ species formed during this process. Rate of Oxidation of Fe2+. As shown in Figure 1, the high-energy line of the quadrupole doublet arising from the Fe2+ species is well resolved from the central doublet, making it possible to accurately estimate the relative absorption intensity of this Fe2+ doublet in samples taken at different reaction time points. The oxidation of Fe2+ as a function of the reaction time can therefore be followed directly by monitoring the decay of the relative intensity of this doublet, since this intensity depends only upon the Fe2+ concentration relative to the total Fe concentration. The results are shown in Figure 2A. The oxidation of the Fe2+ ions is found to be biphasic: a fast and a slow phase. The reaction of apoferritin with Fe2+ and O2 was assumed to have a pseudo-first-order dependence on the Fe2+ concentration, and the data were analyzed by using the following equation
[Fe2+(t)] ) C1 exp(-k1t) + C2 exp(-k2t)
(1)
(Attempts to fit the data with a second-order equation taking into account the dependence on both Fe2+ and oxygen concentrations gave no improvement in the fits.) The solid line plotted in Figure 2A is a least-squares fit to the data using eq 1. The majority of the Fe2+ (C1 ) 72 ( 3%) is oxidized in a fast phase with an apparent pseudo-first-order rate constant k1 of 28.5 ( 3 s-1, with the remainder (C2 ) 28 ( 3%) oxidized at a slower pseudo-first-order rate k2 of 2.3 ( 1 s-1. Characterization of Fe3+ Species Formed during Fe2+ Oxidation by Ferritin. To characterize the various Fe3+ species formed during the oxidation of Fe2+, it is necessary to deconvolute the Mo¨ssbauer spectrum into its spectral components corresponding to the different Fe3+ species. This is made difficult by the fact that these components are only partially resolved, and a unique deconvolution cannot be obtained by using a single spectrum alone. Instead, it is necessary to consider, as a whole, the entire set of Mo¨ssbauer spectra recorded under wide ranges of temperatures (1.5100 K) and applied fields (0-8 T) for all the samples 1 At least two quadrupole doublets are required to properly simulate the line shape of the Fe2+ doublet. The parameters used in the simulation are as follows: ∆EQ(1) ) 3.25 mm/s, δ(1) ) 1.36 mm/s, line width(1) ) 0.43 mm/s, ∆EQ(2) ) 2.72 mm/s, δ(2) ) 1.35 mm/s, and line width(2) ) 0.50 mm/s.
FIGURE 2: Time dependence of all Fe species in the early reaction of H-type subunit apoferritin with Fe2+ and O2, as monitored by Mo¨ssbauer spectroscopy. (A) Decay of the Fe2+ species and (B) formation of the Fe3+ initial products: (b) 0.7 mm/s dimer, (9) 1.2 mm/s dimer, (O) 1.7 mm/s dimer, and (0) trimer. The solid line in panel A is a least-squares fit of eq 1 to the data with a fast rate constant k1 of 28.5 s-1 and a slow rate constant k2 of 2.3 s-1. The solid lines in panel B are theoretical calculations of eq 2 using the same k1 and k2 values (but with varying C1 and C2, see the text), indicating the formation of these ferric species parallels that of the decay of the Fe2+.
quenched at different reaction time points. A self-consistent iterative trial and error approach2 was therefore adopted to obtain a set of spectral components that could satisfactorily explain the entire set of spectra. The results of this analysis are presented below and are represented by the solid lines plotted over the experimental data in Figures 3 and 4 (the contributions from the Fe2+ species have been removed from these spectra). The agreement between the experimental spectra and the simulations based on the analysis is excellent. Due to the complexity of the problem and to avoid confusion, the major findings resulting from the analysis are 2 First, the spectra of the 1 s sample, which contains only 3% ferrous species, were analyzed to obtain initial estimates of the characteristic parameters for the various ferric species. These initial estimates were used to simulate spectra of the ferric species, which were then used for removal of the ferric contribution from the spectra of the 25 ms sample (which contains 55% ferrous species) to obtain spectra representing the ferrous species at various experimental conditions. These prepared spectra of the ferrous species were then used for removing the ferrous contributions from spectra of samples of all other time points to obtain the ferric components for a second analysis for refining the parameters. This entire procedure was then repeated until the data and the spectral simulation converged.
7920 Biochemistry, Vol. 36, No. 25, 1997
FIGURE 3: Mo¨ssbauer spectra at 4.2 K of rapid freeze-quenched samples from the reaction of H-type apoferritin with Fe2+ and O2. These are the same spectra shown in Figure 1 but with the contributions of the Fe2+ species (solid lines in Figure 1) removed and with an expanded velocity scale to enhance the resolution at the central region. Outer regions of the spectra are not shown. The two solid lines plotted at the top of the figure represent the 0.7 and 1.7 mm/s quadrupole doublet, and the dashed line is the 1.2 mm/s quadrupole doublets. The solid lines plotted over the data are superpositions of these three quadrupole doublets with varying intensities, indicating that the central region of these spectra can be explained as a composition of these three doublets.
summarized here first. Four distinct antiferromagnetically coupled Fe3+ species, three dinuclear Fe3+ clusters and one trinuclear Fe3+ cluster, were identified and represented by four different Mo¨ssbauer spectral components. Their characteristic parameters are listed in Table 1. The three dimers can be distinguished by their ∆EQ values and therefore are labeled accordingly as the 1.7, 1.2, and 0.7 mm/s dimers. The trimer shows a ∆EQ value similar to that of the 0.7 mm/s dimer. Very little temperature dependence was found for these ∆EQ values up to 170 K (spectra recorded at a temperature higher than 100 K were collected from a rapidly mixed sample frozen in liquid nitrogen after 2 days of reaction). All four clusters exhibit similar isomer shift value of ∼0.5 mm/s indicative of high-spin ferric compounds with oxygen ligands. A more detailed description of the spectroscopic properties of these ferric-oxy species, which explains the above conclusions, is presented below. To further illustrate the Mo¨ssbauer spectral properties of the ferric-oxy species, we show in Figure 3 the same spectra as those shown in Figure 1 but with the contribution from the ferrous species removed and with an expanded scale near the central region, such that it focuses only on this region for improved resolution and the outer parts of the spectra are not shown. At least three quadrupole doublets are observed with similar δ of ≈0.5 mm/s but with varying ∆EQ values of 0.69, 1.23, and 1.68 mm/s, representing respectively the 0.7, 1.2, and 1.7 mm/s species mentioned above. The 0.7 and 1.7 mm/s doublets are visible as the inner peaks and
Pereira et al.
FIGURE 4: Strong-field (8 T) Mo¨ssbauer spectra at 4.2 K of rapid freeze-quenched samples from the reaction of H-type recombinant frog apoferritin with Fe2+ (36 Fe per 24 apoferritin subunits) and O2. The field is applied parallel to the γ-beam. The samples are the same as those of Figure 1 which were freeze-quenched at (A) 25 ms, (B) 60 ms, (C) 130 ms, (D) 220 ms, (E) 440 ms, and (F) 1 s after rapid mixing. The contributions of the Fe2+ species have been removed from these spectra. The solid lines are superpositions of theoretical simulations of the four initial ferric species using the parameters listed in Table 1 and concentrations listed in Table 2.
outer shoulders, respectively (see the solid lines plotted at the top of Figure 3). The 1.2 mm/s doublet is not easy to visualize, but careful spectral decomposition indicates its presence (dashed line plotted at the top of Figure 3). It is clear from the spectra shown in Figure 3 that all three species are present in samples taken at all the reaction time points. While the isomer shifts of these three doublets are indicative of high-spin Fe3+ (a Kramers system), the fact that the spectra are quadrupole doublets indicates that all three Fe3+containing chemical species share a distinctive electronic property. Ferric quadrupole doublets may arise from complexes with an even number of Fe3+ ions with strong antiferromagnetic exchange coupling to form a diamagnetic ground state. Alternatively, if the electronic relaxation rate is much faster than the nuclear Lamor precession (∼10-7 s for an 57Fe nucleus), the magnetic hyperfine interaction will average out, causing Fe3+ ions in a Kramers system to exhibit a quadrupole doublet. To distinguish ferric quadrupole doublets arising from exchange-coupled diamagnetic ground state and fast electronic relaxation, the spectra were recorded in the presence of a strong magnetic field. Figure 4 shows the spectra of those same samples shown in Figures 1 and 3 recorded at 4.2 K and in the presence of a parallel field of 8 T. In analogy to the weak-field spectra, the three ferric species also absorb at the central region of the strong-field spectra. The solid lines plotted in Figure 4 are simulations using the parameters (∆EQ and δ values) determined from the weak-field spectra and assuming diamagnetism for these three ferric species (i.e., the effective field at the Fe nucleus
Characterization of Ferroxidase Products in H-Ferritin
Biochemistry, Vol. 36, No. 25, 1997 7921
Table 1: Mo¨ssbauer Parameters for the Initial Ferroxidase Products of H-Type Subunit Ferritin species
∆EQ (mm/s)
δ (mm/s)
ηa
A/gnβnb(T)
line width (mm/s)
spin
0.7 mm/s dimer 1.2 mm/s dimer 1.7 mm/s dimer trimer,c site 1 trimer, site 2 trimer, site 3
+0.69 ( 0.04 -1.23 ( 0.04 -1.68 ( 0.04 +0.7 ( 0.1 -0.7 ( 0.1 +0.7 ( 0.1
0.53 ( 0.02 0.50 ( 0.02 0.51 ( 0.02 0.51 ( 0.04 0.51 ( 0.04 0.51 ( 0.04
1.0 ( 0.4 1.1 ( 0.4 0.6 ( 0.4 -0.7 ( 0.2 -1.0 ( 0.2 -0.2 ( 0.2
0 0 0 -22.6 ( 0.3 +17.4 ( 0.3 -19.1 ( 0.3
0.35 0.32 0.30 0.5 0.5 0.5
0 0 0 5 /2 5 /2 5 /2
a η ) (Vxx - Vyy)/Vzz is the asymmetry parameter, where Vii are the principal components of the electric field gradient at the Fe nucleus. b A is the magnetic hyperfine coupling constant, and gnβn is the magnetic moment of the 57Fe nucleus. c The trimer is assumed to have a rhombicity parameter E/D of 1/3.
of these species is equal to the applied field). The agreement between experiment and simulations at the central region of the spectra provides definitive evidence for the diamagnetism of these three ferric species. This observed diamagnetism indicates that these ferric species are most probably antiferromagnetically coupled (Fe3+)2-oxy clusters. Since a dinuclear Fe3+ cluster may contain iron sites with different ligand environments, it is possible for a single dinuclear cluster to exhibit two quadrupole doublets as observed in the R2 subunit of the E. coli ribonucleotide reductase (Lynch et al., 1989). A comment is therefore in order here to address the question of whether these doublets observed in ferritin may represent different iron sites of a single dinuclear cluster. A necessary criterion for assigning two quadrupole doublets to two iron sites of a dinuclear iron center is that the two doublets must have the same intensity, which is the case for ribonucleotide reductase. Since the three doublets observed in ferritin have different intensities, they cannot represent signals arising from different iron sites of a single dinuclear center. Furthermore, this difference in intensity also cannot be explained as being due to different occupations of the iron sites when coupled with the observed diamagnetism of the doublets, because only a fully occupied diiron center would exhibit diamagnetism. A monomeric Fe3+ (a diiron center lacking one iron), on the other hand, is paramagnetic. Consequently, the only possible assignment for the three quadrupole doublets observed in ferritin is that they represent three distinct (Fe3+)2-oxy clusters. The Mo¨ssbauer spectral properties of the Fe3+ trimer are more complex. Its electronic relaxation is neither very fast nor slow on the time scale of the nuclear Larmor precession frequency, resulting in the broad and ill-defined magnetic spectral component observed in the 4.2 K weak-field spectra (Figure 1). This makes spectral analysis difficult. The problem can be overcome by collecting the spectra under strong applied fields for at fields of sufficient strength (i.e., strong enough so that the magnetically purturbed electronic ground state becomes stabilized by an energy of >kT relative to the excited states) the spectra of a slow-relaxing species and a fast-relaxing species converge to the same spectral profile. We have therefore collected spectra with varying applied fields. With increasing field strengths, the broad and unresolved magnetic features sharpen into recognizable lines as seen in the outer region of the spectra shown in Figure 4. Figure 5 shows the magnetic spectral component corresponding to the Fe3+ trimer recorded at 4, 6, and 8 T applied fields. These spectra were prepared from those of the 1 s sample by removing the contributions from the ferrous species and the three Fe3+ dimers. The 1 s sample was used primarily because it contains the least amount of ferrous species, of which the spectral properties as a function of applied fields
FIGURE 5: Field-dependent Mo¨ssbauer spectra of the trimer recorded at 4.2 K. The experimental spectra are prepared from spectra of the 1 s sample recorded with a magnetic field of (A) 4 T, (B) 6 T, or (C) 8 T applied parallel to the γ-beam. The solid lines plotted over the experimental spectra are superpositions of theoretical spectra of the three Fe sites of the trimer simulated with the parameters listed in Table 1. Simulations for the individual Fe sites are also plotted above each corresponding experimental spectrum: sites 1 and 3, solid lines; and site 2, dashed line.
are least understood. Consequently, any “possible” distortion of the prepared magnetic spectrum caused by “improper” removal of the ferrous contribution is minimized by the use of the 1 s sample. The spectral behavior of the three Fe3+ dimers as a function of the applied field can be accurately simulated because they are diamagnetic and because their ∆EQ and δ values have been determined from the weakfield spectra. The spectra shown in Figure 5 can be explained as superpositions of three magnetic spectral components of equal intensity corresponding to three Fe3+ sites of an exchange-coupled cluster (i.e., a trinuclear Fe cluster). The cluster spin is assumed to be 5/2 as supported by the g ) 4.3 resonance in the EPR spectrum discussed below. The antiferromagnetically coupled nature of the cluster is indicated unambiguously by the variation of the total magnetic splittings of the component spectra with applied field strength. The total splittings for two of the three components (solid lines plotted above the spectra in Figure 5) decrease with increasing applied field strengths. This behavior is generally expected for an 57Fe nucleus, since the magnetic hyperfine constant A for Fe ions is negative,
7922 Biochemistry, Vol. 36, No. 25, 1997 i.e., the strong internal hyperfine field opposes the weaker applied field, resulting in the decrease of the effective field at the Fe nucleus with increasing applied field strengths. However, the splitting of the third component (dashed lines plotted above the spectra of Figure 5) increases with increasing applied field strengths. This could happen only if the corresponding Fe atom is antiferromagnetically coupled to other Fe atoms such that its intrinsic spin is antiparallel to the system spin, resulting in an apparent positive magnetic hyperfine A value (Huynh et al., 1980; Kent et al., 1980). Since the field-dependent behavior of one of the magnetic components indicates the presence of antiferromagnetic coupling and since coupling of two ferric ions would not yield a half-integer spin system, all three iron sites corresponding to the three magnetic components must be involved in the exchange coupling. In other words, the three magnetic spectral components must represent three iron sites of an exchange-coupled cluster, i.e., a trinuclear Fe3+cluster. Inspection of the outer regions of the spectra shown in Figure 4 indicates clearly that this trimer is present in all the samples quenched at different reaction times. Rates of Formation of the Fe3+ Species Resulting from Fe2+ Oxidation by Ferritin. The analysis described above provides an accurate estimate of the relative absorption intensity of each of the four ferric species in the samples in addition to yielding information concerning the nuclearity and electronic properties of the ferric clusters. The formation of these Fe3+ species as a function of the reaction time can therefore be obtained with the results shown in Figure 2B. Similar to the oxidation of the Fe2+ species, the formation of these four Fe3+ clusters is also biphasic. The data suggest that all four species are formed simultaneously, instead of sequentially, with rates similar to those of the decay of the Fe2+ species. To illustrate this point, formation curves (solid lines) for the four species are plotted in Figure 2B using eq 2 and rate constants (k1 ) 28.5 s-1 and k2 ) 2.3 s-1) determined for the decay of the Fe2+ species.
C(t) ) C1[1 - exp(-k1t)] + C2[1 - exp(-k2t)] (2) Parameter C1 and C2 used in the calculations for the 1.7 mm/s dimer, 1.2 mm/s dimer, 0.7 mm/s dimer, and trimer are, respectively, 9.6 and 7.3%, 10 and 1.2%, 20 and 6%, and 28.8 and 14.8%. The agreement between the calculations and the experimental data demonstrates that within experimental uncertainties the four Fe3+ clusters form in parallel with the decay of the Fe2+ species. In other words, multiple processes are involved immediately following the oxidation of ferrous ions by ferritin. Multiple Fe3+ species, similar to those reported here (see Discussion), have previously been observed 0.5, 10, or 20 min after mixing Fe2+ with human recombinant H-type subunit ferritin in the presence of O2 (Bauminger et al., 1989, 1991, 1993). The ferric species identified in the studies of recombinant human ferritin have been interpreted to show a precursor/product relationship (Bauminger et al., 1993). The field-dependent Mo¨ssbauer data on rapid freeze-quench samples presented here, however, show clearly that these ferric species form simultaneously, at least in frog H-type subunit ferritin, and that the multiple oxidation products occur as early as 25 ms after mixing Fe2+ and protein. EPR spectroscopy of Rapidly Forming Species in HSubunit-Type Ferritins. The rapid freeze-quench method was
Pereira et al.
FIGURE 6: Comparison of the time course of the EPR g ) 4.3 signal intensity (filled squares) with the formation of the trimer (empty circles) observed by Mo¨ssbauer spectroscopy. The EPR data were recorded at 8 K with a microwave power of 2 mW (see Methods for other experimental conditions). The Mo¨ssbauer data and the solid curve are the same of those presented in Figure 2.
used to prepare samples for EPR investigations in parallel with the samples for Mo¨ssbauer measurements in order to explore more completely the properties of the paramagnetic species observed in the Mo¨ssbauer measurements. Samples were made with 2.4 mM Fe and the same 36/1 Fe/ferritin ratio. The samples were quenched at reaction times of 0.025, 0.045, 0.10, 0.25, 0.44, and 1 s. Two EPR signals were detected at low temperature (8 K), one at g ) 4.3 and one at g ) 2.0 (data not shown). The shape of the EPR signal at the g ) 2.0 region suggests a protein-based radical. A similar signal has been observed in human recombinant H-type subunit ferritin and has been reported previously (Yu et al., 1995). The g ) 4.3 signal is typical of mononuclear Fe3+ in a low-symmetry environment with an E/D of 1/3 and an S of 5/2. However, an exchange-coupled system with an odd number of Fe3+ ions could also form a ground state with an S of 5/2. Since the Mo¨ssbauer data presented above show only one paramagnetic species, namely, the trimer, there is no other alternative but to assign the g ) 4.3 signal to the trimer. Absolute quantification of the g ) 4.3 signal was not attempted since the signal arises from an excited state of the S ) 5/2 multiplet, making its quantitation difficult without a priori knowledge of the multiplet fine structure splittings. Nevertheless, the relative intensity of the g ) 4.3 EPR signal as a function of reaction time can be compared with the time dependence of the trimer formation obtained from the Mo¨ssbauer measurements (Figure 6). Although the EPR data show a large scattering (probably due to variation in the sample density), the comparison of EPR and Mo¨ssbauer data suggests that the time course of the g ) 4.3 EPR signal intensity conforms with the formation of the trimer.3 Formation of the Polynuclear Fe3+ Cluster by H-Type Subunit Ferritin at 36 Fe/Protein. To investigate whether H-type subunit ferritin is capable of forming a large polynuclear ferrihydrite mineral core at small Fe loading and to examine whether the above-mentioned small ferric species are transient during ferritin mineralization, apoferritin was 3 An EPR sample was prepared with apoferritin rapidly mixed with Fe2+ and O2 and allowed to react aerobically for 2 days. The sample exhibits the same EPR signal at g ) 4.3 with a reduction in intensity consistent with the decrease of the amount of the trimer determined from Mo¨ssbauer measurements of the Mo¨ssbauer 2-day-old sample.
Characterization of Ferroxidase Products in H-Ferritin
FIGURE 7: Mo¨ssbauer spectrum of a sample of H-type frog apoferritin rapidly mixed with 36 equiv of Fe2+ in the presence of O2 and frozen 2 days after mixing. The spectrum (vertical bars) was recorded at 4.2 K with an 8 T magnetic field applied parallel to the γ-beam. The dashed line shows the spectrum of the young mineral at 8 T (see the text for the preparation of this spectrum).
rapidly mixed with Fe2+ (36 Fe/protein) in the presence of O2 and allowed to react aerobically for 2 days at 4 °C before freezing at 77 K for Mo¨ssbauer measurements. Figure 7 shows a Mo¨ssbauer spectrum of the 2-day-old sample recorded at 4.2 K with a parallel applied field of 8 T. Spectral components associated with the above-mentioned Fe3+ dimers and trimer are clearly observable. However, their relative absorption intensities are reduced in comparison with those of the 1 s sample (see Table 2). Except for the 0.7 mm/s dimer, the reduced levels of the other species are outside of the experimental errors. Removal of the contributions of these small ferric clusters reveals a complex magnetic spectrum (dashed line in Figure 7, representing ≈32% of the total Fe absorption or approximately 12 Fe/ferritin on average) that can be attributed to larger (n > 3) polynuclear ferric clusters. A typical ferritin mineral would have detectable superparamagnetism (St. Pierre et al., 1986). Even as few as 10 Fe/ferritin formed a readily detectable superparamagnetic ferric species in an L-subunit-type ferritin (Yang et al., 1987). In order to demonstrate that superparamagnetic species are indeed present in the 2-day-old H-subunit ferritin sample, spectra were recorded at various temperatures with a weak or in the absence of an applied field. The 40 and 4.2 K spectra of the 2-day-old sample are shown in Figure 8 (spectra A and B, respectively). In addition to the spectral components of the four small clusters (the 0.7, 1.2-, and 1.7 mm/s dimers and the trimer) seen at shorter reaction times, a broad magnetic sextet spectral component is visible in the 4.2 K spectrum (Figure 8B), suggesting the presence of larger clusters. Since the relative concentrations of the small initial forming clusters in the 2-day-old sample can be determined from the 8 T spectrum (Figure 7), and since the spectral properties of these small clusters can be obtained from the spectra of the 1 s sample recorded under the same experimental conditions, their contributions can be removed from the 4.2 K spectrum of the 2-day-old sample to reveal the spectral component corresponding to the larger clusters. Figure 8C shows such a prepared spectrum with a broad sextet magnetic pattern displaying a distribution of internal magnetic field of 45-46 T. At 40 K, however, this sextet spectrum is not visible and has collapsed into a quadrupole doublet that is unresolved from the other quadrupole doublets (Figure 8A). Such a temperature dependence is typical for
Biochemistry, Vol. 36, No. 25, 1997 7923 superparamagnetic systems. Due to the presence of multiple species in the 2-day-old sample and the relatively small contribution of the superparamagnetic species, the blocking temperature4 of the superparamagnetic species could not be determined. However, the fact that the spectrum collapses at 40 K indicates that these superparamagnetic clusters have a blocking temperature much lower than that of a mature ferritin mineral core; the blocking temperature of mature horse spleen ferritin reconstituted with 480 Fe/protein had been determined to be 38 K (Yang et al., 1987). Since the blocking temperature is proportional to the average size of the superparamagnetic particles, the observed lower blocking temperature suggests that these superparamagnetic clusters in the 2-day-old sample are much smaller than the mature ferritin core, and may represent polynuclear ferric clusters formed in the early stage of ferritin mineralization. Thus, these clusters are designated the young ferritin core and are characterized by a blocking temperature of below 40 K and a distribution of internal fields at 45-46 T. DISCUSSION The first step in the formation of the iron mineral in ferritin is oxidation of Fe2+ to Fe3+. Mainly indirect methods have been used until recently to analyze rates of ferroxidation by ferritin, especially for fast (H-subunit-type) ferritin. Such methods have included measurement of the decrease in the formation of the ferrous 1,10-phenanthroline complex (Macara & Harrison, 1973), appearence of polynuclear ferricoxo species (absorbance in the 330-420 nm range) [e.g., Bryce and Crichton (1973), Macara et al. (1973), and Mertz and Theil (1983)], acceleration of formation of ferrictransferrin in acetate buffer (Baker & Boyer, 1986), formation of purple ferric-tyrosine (Waldo et al., 1993), and generation of EPR active ferric complexes (Hanna et al., 1991; Sun & Chasteen, 1994). Besides being indirect, additional limitations exist for these methods. In the case of using the chelator 1,10-phenanthroline, the method does not always distinguish the oxidation of Fe2+ from the inaccessible Fe2+ [e.g., Rohrer et al. (1989)]. EPR spectroscopy is limited by the inability to detect all the ferric species present such as the diamagnetic Fe3+ dimers observed in this work. In general, methods depending on the appearance of Fe3+ in a particular environment are limited by the necessary assumption that the species measured is the initial oxidation product, which is not necessarily the case. Thus, while the ferrictyrosine forms rapidly (Waldo & Theil, 1993), for example, it could be a product of an even faster forming colorless ferric species that is the initial ferroxidation product. Here, we adopted an approach that combined the rapid freeze-quench method and Mo¨ssbauer spectroscopy to investigate in detail the oxidation of Fe2+ ions by H-type subunit ferritin. Application of the rapid freeze-quench method allowed us to quench the ferroxidation reaction at various time points along the reaction pathway for spectroscopic investigations. The time resolution of the rapid freeze-quench method is in the millisecond range, much shorter than what was possible in previous Mo¨ssbauer measurements (Yang et al., 1987; Bauminger et al., 1989, 1991, 1993). A major advantage of using Mo¨ssbauer 4 The blocking temperature is defined as the temperature where half of the iron absorption is in a magnetic sextet and half is in a quadrupole doublet.
7924 Biochemistry, Vol. 36, No. 25, 1997
Pereira et al.
Table 2: Percentage of the Total Fe Absorption of the Ferric Species Formed at Various Reaction Times, tR, during Mineralization of H-Type Subunit Ferritina
a
tR
0.7 mm/s dimer
1.2 mm/s dimer
1.7 mm/s dimer
trimer
young mineral
25 ms 60 ms 130 ms 220 ms 440 ms 1s 48 h
10 (8-11) 19 (17-20) 21 (20-22) 22 (21-23) 23 (21-24) 26 (25-27) 24 (23-28)
8 (7-10) 8 (7-10) 10 (9-12) 11 (10-12) 11 (9-12) 9 (7-11) 6 (3-7)
7 (4-9) 8 (7-10) 10 (9-11) 13 (12-14) 14 (13-16) 16 (13-17) 9 (8-10)
12 (10-14) 24 (22-25) 32 (30-33) 35 (34-37) 37 (36-39) 42 (41-43) 29 (22-30)
0 0 0 0 0 3), is more complicated. Examination of the species with a ∆EQ of 0.7 mm/s at strong magnetic fields (4-8 T) reveals the presence of two species, one diamagnetic and one paramagnetic, an unexpected finding according to conclusions made previously on the basis of data obtained at weak fields (Bauminger et al., 1989, 1991, 1993). The diamagnetism of one of these species strongly suggests that this species is yet another antiferromagnetically coupled dinuclear Fe3+ cluster. Strongfield Mo¨ssbauer spectra further reveal that the paramagnetic species is composed of three antiferromagnetically coupled Fe3+ ions. Correlating the Mo¨ssbauer results with EPR measurements on samples prepared under similar conditions suggests that the Fe3+ trimer has a system spin of 5/2 and exhibits EPR signals indistinguishable from that of a monomeric Fe3+ species. In fact, such an EPR signal with a formation rate very similar to that of the trimer has been reported during the reconstitution of recombinant human H-type subunit ferritin and was attributed to monomeric ferric species (Sun & Chasteen, 1994). However, on the basis of our current findings, it is very likely that the EPR signal detected in the human H-type subunit ferritin during the early stage of ferroxidation reaction may also represent a Fe3+ trimer similar to that observed in frog ferritin. A trimeric Fe3+-oxy model complex exhibiting a typical S ) 5/2, g ) 4.3 EPR signal and magnetic hyperfine coupling constants similar to those of the trimer reported here has been reported recently (Bill et al., 1997). The ∆EQ values of µ-oxo diferric clusters are typically large, within the range of 1.4-2.4 mm/s (Kurtz, 1990; Que & True, 1990). The small ∆EQ of 0.7 mm/s found for the diamagnetic dinuclear Fe3+ species formed in ferritin may indicate a µ-hydroxo cluster (Kurtz, 1990) or represent a local Fe environment resembling that in the ferritin mineral core which exhibits a similar ∆EQ value (St. Pierre et al., 1986; Yang et al., 1987). A similar quadrupole doublet has been observed previously in horse spleen ferritin and recombinant human H-type mutants and was assigned as small superparamagnetic ferric clusters (Bauminger et al., 1991, 1993). Since the samples examined in these studies were not prepared with rapid mixing-rapid quench techniques and were frozen at reaction times (minutes to hours) much longer than those of the current study (except for the 2-day-old sample), it is possible that small superparamagnetic ferric clusters have been formed. However, in view of our current finding that various Fe3+ species in ferritin (the 0.7 mm/s dimer, trimer, and mineral core) exhibit a similar 0.7 mm/s quadrupole doublet, it is very likely that the doublet observed in these previous studies could represent a mixture of species, particularly in samples taken at the early reaction time points. The parallel formation of all four ferric oxy clusters coincident with the oxidation of the Fe2+ ions is a unique observation of the current study. Their formations are all biphasic with a fast and a slow rate similar to that of the decay of the Fe2+ ions. This observation strongly suggests that multiple processes are involved following the oxidation of ferrous ions by H-type subunit ferritin and that all four ferric species are initial oxidation products. At first glance,
Biochemistry, Vol. 36, No. 25, 1997 7925 this observation of multiple ferric species appears to be at odds with the studies of X-ray crystallography (Lawson et al., 1991; Hempstead et al., 1994) and of site-directed mutants of H-subunit-type ferritin (Sun et al., 1993; Yu et al., 1995; Harrison & Arosio, 1996), which suggest that fast Fe2+ oxidation occurs at a single proposed diiron binding site. However, it is important to point out that ferritin, as the only protein which directs the conversion of ferrous ions in solution into forming the solid iron mineral core in the ferritin cavity, has unique functional features. Our results indicate that immediately following the ferrous oxidation several types of multimeric ferric species form exceedingly fast and at the same rates. The results neither support nor refute the presence of a single ferroxidation site. It is not yet clear which sites are the sites for the multiple initial oxidation products. However, it is interesting to note that mutation of Y305 eliminates several of these Fe3+-oxy species (Bauminger et al., 1993; also, see below) and that Y30 is a conserved residue near the proposed dinuclear ferroxidation center. If there were indeed a single ferroxidation site in each H-type ferritin subunit, as suggested by the work of X-ray crystallography (Lawson et al., 1991; Hempstead et al., 1994) and of site-directed mutants (Sun et al., 1993; Yu et al., 1995; Harrison & Arosio, 1996), our results may reflect the flexibility of the site since it is able to accommodate various initial oxidation products. Considering the ferritin function, this flexibility is not surprising and may be important since these initial oxidation products would have to be eventually translocated into the protein cavity. The plasticity of ferritin has been noted in a previous protein crystallography study (Trikha et al., 1995). There are several possible explanations for why the oxidation is observed to be biphasic. One explanation would be that there are two subtypes of ferritin in our samples such that one reacts faster and the other slower. In order for this hypothesis to explain the observed biphasic curve in these samples, however, the Fe2+ must bind tightly to the “slow” ferritin, for otherwise the “fast” ferritin would oxidize essentially all the Fe2+ ions within ≈0.2 s. Also, in order to explain the parallel formation of the four oxidation products, both the fast and slow ferritin molecules would have to catalyze the oxidation reaction with the same four parallel processes. This appears to be rather unlikely. A more plausible explanation based on ferritin function is a classical burst reaction (Bender et al., 1966) which assumes a fast initial step followed by a much slower rate-limiting conversion of the initial product into the final product. In the case of ferritin, the slow step could be the conversion of these initially forming small ferric clusters into a solid phase mineral core, which involves transporting and recombining these clusters into larger polynuclear clusters. Such an explanation is supported by the measurements on the 2-dayold samples which indicate that larger polynuclear clusters have been formed at the expense of the smaller initially forming clusters (Table 2). The similarity of the rate of formation of ferric-tyrosinate species (920 s-1) detected by optical spectroscopy (Waldo & Theil, 1993) with the rate of oxidation of the Fe2+ (1026 5 Two numbering systems have been used for ferritin sequences. In the numbering system introduced by Lawson et al. (1991), 4 is added to the original sequence assignement. Thus, Y30F is equal to Y34F in Bauminger et al. (1993).
7926 Biochemistry, Vol. 36, No. 25, 1997 s-1) determined here indicates that the ferric-tyrosinate species is an initial oxidation product. The current Mo¨ssbauer measurements indicate that there are four initial oxidation products in the form of small ferric clusters. Taken together, these observations suggest that there is a high probability that more than one, if not all, of the small ferricoxy clusters are coordinated to protein tyrosine residues. This assignment of the small Fe3+-oxy species to the ferrictyrosine species, which show an absorption maximum at 550 nm, is consistent with that observed for purple acid phosphatases which exhibit an absorption maximum between 550 and 570 nm (Doi et al., 1988). Spectroscopic investigations of rapid freeze-quenched samples of site specific mutants of conserved tyrosine residues should help to elucidate the location of the sites of these initial Fe3+-oxy-tyrosinate clusters. Earlier Mo¨ssbauer studies on recombinant human H-type subunit ferritin (Bauminger et al., 1993) showed that substitution of tyrosine 30 for phenylalanine (Y30F) eliminated the 0.7 and 1.62 mm/s (corresponding to the 1.7 mm/s doublet in frog ferritin) doublets in samples prepared 0.5 and 30 min after mixing of Fe2+ with protein. Note that the 0.7 mm/s doublet was assigned to larger (n > 3) ferric clusters. But, since strong-field measurements needed to fully characterize the Fe3+ species were not performed, it is probable that a mixture of ferroxidase products (i.e., the 0.7 mm/s dimer, the trimer, and the young ferritin core) may be contributing to the 0.7 mm/s doublet. Thus, assuming that the 0.7 mm/s dimer and the trimer are also part of the initial ferroxidase products in the human H-subunit ferritin, the reported Mo¨ssbauer data on the Y30F mutant, which show an absence of the 1.62 and 0.7 mm/s doublets at 1 min of reaction time (Bauminger et al., 1993), suggested that Y30 is involved in the formation of the trimer, the 0.7 mm/s dimer, and the 1.7 mm/s dimer. Stopped-flow UV-vis studies showed that Y30F substitution slows the rates of mineralization in human H-subunit ferritin (Treffrey et al., 1995) and of ferric-tyrosine formation in frog H-subunit ferritin (Fetter, Cohen, Sander-Loehr, and Theil, to be published). These observations can be explained by our finding that multiple processes are taking place following ferritin ferroxidation and by the suggestion that residue Y30 is involved in most but not all the processes. Multiple postferroxidation processes occurring at different protein sites (or a flexible site) also explain previous results of sitedirected mutagensis of human H-subunit ferritin which altered the rate of ferroxidation but did not inhibit ferritin function completely (Sun et al., 1993; Bauminger et al., 1993; Treffry et al., 1995). Multiple parallel oxidation pathways in H-subunit ferritin provide an alternate model in contrast to that previously proposed (Bauminger et al., 1989, 1991, 1993). The earlier model, based on Mo¨ssbauer data collected at limited and longer time points (basically 0.5 and 30 min; Bauminger et al., 1993) and on site-directed mutations at the 3-fold channel (Treffry et al., 1993), suggested that ferrous ions are to be oxidized at a single ferroxidase site in the interior of the four-helix bundle to form dinuclear Fe3+-oxo clusters (i.e., the 1.2 and 1.7 mm/s dimers). The Fe3+-oxo dimers are split into monomers, which are then transported from the ferroxidation site to the 3-fold channels and then to the cavity in the center of the protein to form the solid ferritin mineral core. The proposed initial formation of the Fe3+-oxo dimers is supported by our rapid freeze-quench Mo¨ssbauer data, but
Pereira et al. the proposed subsequent event of spliting the dimers into monomers appears to be energetically unfavorable since dinuclear Fe3+-oxo clusters are stable compounds and the proposal is inconsistent with our data which show an absence of monomeric Fe3+ species. Our data suggest that the initial postoxidation of Fe2+ by H-ferritin involves multiple processes and that the final formation of the ferritin mineral requires transporting, followed by combining the initially formed Fe3+-oxy clusters, because (1) the formation of the trimer and various Fe3+ dimers occurs in parallel with the oxidation of Fe2+ ions, (2) no monomeric ferric species were observed in our rapid freeze-quench samples and the 2-dayold sample, (3) the observed biphasic behavior of the ferrous oxidation and initial cluster formation is consistent with fast formation of initial products at one or more protein sites followed by a slow process of transporting the initial products to the nucleation site(s), (4) fast ferroxidase activity is restored very slowly after the addition of the initial group of Fe2+ ions (Waldo & Theil, 1993; Treffry et al., 1995), which reflects the slow process of vacating the ferroxidase and/or the postoxidation sites, and (5) the young ferritin mineral core is formed at the expense of the initially forming Fe3+-oxy clusters. Finally, using small ferric-oxy clusters as building blocks to form larger clusters is energetically possible and has been observed in model compound studies (Gorun & Lippard, 1986; Nair & Hagen, 1992; Taft et al., 1993; Caneschi et al., 1995). The multiplicity of ferric clusters which form in the initial stages of mineralization, as observed by rapid freeze-quench Mo¨ssbauer spectroscopy, emphasizes the relatively unspecific Fe-protein interactions in ferritin compared to proteins in which the metal is part of a catalytic center required for function. In ferritin, the metal is a substrate for oxidation and transport. Whether each of the 24 subunits in ferritin has “active sites” for binding, oxidizing, and/or transporting Fe ions across the protein to and from the cavity, or the sites are formed in regions between neighboring subunits, or a single subunit may have multiple types of metal binding sites cannot be determined at the current level of knowledge; on the basis of the kinetic Mo¨ssbauer data presented here for the multiple ferric species observed with 36 Fe added per protein, there are an average of 10-12 sites per ferritin molecule (24 subunits). There appears to be no strong evolutionary pressure on ferritin for a single process for Fe oxidation and transfer across the protein. Rather, the main requirement for ferritin mineralization may be protein folding and packing which produce an accessible interior hollow in which formation of a solid phase mineral core can occur. Illustrations of “minimalist” ferritin structures include (1) a quadruple Glu/Ala-mutated L-subunit ferritin which folds and assembles correctly (Trihka et al., 1994, 1995) but lacks protein-based ferroxidase or nucleation sites (Wade et al., 1991; Trihka et al., 1995; Waldo & Theil, 1996) and (2) a form of concanavalin A which assembles with an interior hollow in the protein (Yariv et al., 1988). Both proteins form a polynuclear ferric-oxo species in Vitro. Evolutionary nuances, crucial to biology, that are indicated by the specificity of genetic control of expression of H and L subunits apparently include the acquisition of Fe transport paths in ferritin with increasing speed, but the protein assembly to form a commodious interior space, common to both H- and L-subunit ferritin, clearly dominates.
Characterization of Ferroxidase Products in H-Ferritin ACKNOWLEDGMENT The authors are grateful to Dr. Geoffrey S. Waldo for his contributions to the initial experiments of this work and to Mr. Joel C. Verbist for his assistance in collecting some of the strong-field Mo¨ssbauer spectra. REFERENCES Baker, G. R., & Boyer, R. F. (1986) J. Biol. Chem. 261, 1318213185. Balla, G., Jacob, H. S., Balla, J., Rosenberg, M., Nath, K., Apple, F., Eaton, J. W., & Vercellotti, G. M. (1992) J. Biol. Chem. 267, 18148-18153. Banyard, S. H., Stammers, D. K., & Harrison, P. M. (1978) Nature 271, 282-284. Bauminger, E. R., Harrison, P. M., Nowik, I., & Treffry, A. (1989) Biochemistry 28, 5486-5493. Bauminger, E. R., Harrison, P. M., Hechel, D., Nowik, I., & Treffry, A. (1991) Proc. R. Soc. London, Ser. B 244, 211-217. Bauminger, E. R., Harrison, P. M., Hechel, P. M., Hodgson, N. W., Nowik, I., Treffry, A., & Yewdall, S. (1993) Biochem. J. 296, 709-719. Bender, M. L., Begue´-Canto´n, M. L., Blakeley, R. L., Brubacher, L. J., Feder, J., Gunter, C. R., Ke´zdy, F. J., Killheffer, J. V., Jr., Marshall, T. H., Miller, C. G., Roeske, R. W., & Stoops, J. K. (1966) J. Am. Chem. Soc. 88, 5890-5913. Bill, E., Krebs, C., Winter, M., Gerdan, M., Trautwein, A. X., Flo¨rke, U., Haupt, H.-J., & Chaudhuri, P. (1997) Chem. Eur. J. 3, 193-201. Blackwell, J. R., & Horgan, R. A. (1991) FEBS Lett. 295, 10-12. Bryce, C. F. A., & Crichton, R. R. (1973) Biochem. J. 133, 301309. Caneschi, A., Cornia, A., Fabretti, A. C., & Gatteschi, D. (1995) Angew. Chem., Int. Ed. Engl. 34, 2716-2718. Dean, J. A. (1992) Lang’s Handbook of Chemistry, p 57, McGrawHill, New York. DeWitt, J. G., Bentsen, J. G., Rosenzweig, A. C., Hedman, B., Green, J., Pilkington, S., Papaefthymiou, G. C., Dalton, H., Hodgson, K. O., & Lippard, S. J. (1991) J. Am. Chem. Soc. 113, 9219-9235. Didsbury, J. R., Theil, E. C., Kaufman, R. E., & Dickey, L. F. (1986) J. Biol. Chem. 261, 949-955. Doi, K., Antanaitis, B. C., & Aisen, P. (1988) Struct. Bonding (Berlin) 70, 3-26. Fox, B. G., Hendrich, M. P., Surerus, K. K., Andersson, K. K., Froland, W. A., Lipscomb, J. D., & Mu¨nck, E. (1993) J. Am. Chem. Soc. 115, 3688-3701. Gorun, S. M., & Lippard, S. J. (1986) Nature 319, 666-668. Hanna, P. M., Chen, Y., & Chasteen, N. D. (1991) J. Biol. Chem. 266, 886-893. Harrison, P. M., & Lilley, T. H. (1990) in Iron Carriers and Iron Proteins (Loehr, T. M., Ed.) pp 353-452, VCH, New York. Harrison, P. M., & Arosio, P. (1996) Biochim. Biophys. Acta 1275, 161-203. Hempstead, P. D., Hudson, A. J., Artymiuk, P. J., Andrews, S. C., Banfield, M. J., Guest, J. R., & Harrison, P. M. (1994) FEBS Lett. 350, 258-262. Huynh, B. H. (1994) Methods Enzymol. 243, 523-543. Huynh, B. H., Moura, J. J. G., Moura, I., Kent, T. A., LeGall, J., Xavier, A. V., & Mu¨nck, E. (1980) J. Biol. Chem. 255, 32423244. Kent, T. A., Huynh, B. H., & Mu¨nck, E. (1980) Proc. Natl. Acad. Sci. U.S.A. 77, 6574-6576.
Biochemistry, Vol. 36, No. 25, 1997 7927 Kurtz, D. M., Jr. (1990) Chem. ReV. 90, 585-606. Lawson, D. M., Artymiuk, P. J., Yewdall, S. J., Smith, J. M. A., Livingstone, J. C., Treffry, A., Luzzago, A., Levi, S., Arosio, P., Cesareni, G., Thomas, C. D., Shaw, W. V., & Harrison, P. M. (1991) Nature 349, 541-544. LeBrun, N. E. L., Cheesman, M. R., Thomson, A. J., Moore, G. R., Andrews, S. C., Guest, J. R., & Harrison, P. M. (1993) FEBS Lett. 323, 261-266. Lynch, J. B., Juarez-Garcia, C., Mu¨nck, E., & Que, L., Jr. (1989) J. Biol. Chem. 264, 8091-8096. Macara, I. G., Hoy, T. G., & Harrison, P. M. (1973) Biochem. J. 135, 343-348. Mertz, J. R., & Theil, E. C. (1983) J. Biol. Chem. 258, 1171911726. Nair, V. S., & Hagen, K. S. (1992) Inorg. Chem. 31, 4048-4050. Nordlund, P., Sjo¨berg, B. M., & Eklund, H. (1990) Nature 345, 593-598. Que, L., Jr., & True, A. E. (1990) in Progess in Inorganic Chemistry: Bioinorganic Chemistry (Lippard, S. J., Ed.) Vol. 38, pp 97-200, John Wiley & Soms, Inc., New York. Ravi, N., Bollinger, J. M., Jr., Huynh, B. H., Edmondson, D. E., & Stubbe, J. (1994) J. Am. Chem. Soc. 116, 8007-8014. Rohrer, J. S., Frankel, R. B., Papaefthymiou, G. C., & Theil, E. C. (1989) Inorg. Chem. 28, 3393-3395. Rosenweig, A. C., Frederick, C. A., Lippard, S. J., & Nordlund, P. (1993) Nature 366, 537-543. St. Pierre, T. G., Bell, S. H., Dickson, D. P. E., Mann, S., Webb, J., Moore, G. R., & Williams, R. J. P. (1986) Biochim. Biophys. Acta 870, 127-134. Sun, S., & Chasteen, N. D. (1994) Biochemistry 33, 15095-15102. Sun, S., Arosio, P., Levi, S., & Chasteen, N. D. (1993) Biochemistry 32, 9362-9369. Taft, K. L., Papaefthymiou, G. C., & Lippard, S. J. (1993) Science 259, 1302-1305. Theil, E. C. (1987) Annu. ReV. Biochem. 56, 289-315. Touati, D., Jacques, M., Tardat, B., Bouchard, L., & Despied, S. (1995) J. Bacteriol. 177, 2305-2314. Treffry, A., Bauminger, E. R., Hechel, D., Hodson, N. W., Nowik, I., Yewdall, S. J., & Harrison, P. M. (1993) Biochem. J. 296, 721-728. Treffry, A., Zhao, Z., Quail, M. A., Guest, J. R., & Harrison. P. M. (1995) J. Biochemistry 34, 15204-15213. Trihka, J., Waldo, G. S., Lewandowski, F. A., Theil, E. C., Weber, P. C., & Allewell, N. M. (1994) Proteins 18, 107-118. Trihka, J., Theil, E. C., & Allewell, N. M. (1995) J. Mol. Biol. 248, 949-967. Wade, V. J., Levi, S., Arosio, P., Treffry, A., Harrison, P. M., & Mann, S. (1991) J. Mol. Biol. 221, 1443-1452. Waldo, G. S., & Theil, E. C. (1993) Biochemistry 32, 1326113269. Waldo, G. S., & Theil, E. C. (1996) in ComprehensiVe Supramolecular Chemistry (Suslick, K. S., Ed.) Vol. 5, pp 65-89, Pergamon Press, Oxford, UK. Waldo, G. S., Ling, J., Sanders-Loehr, J., & Theil, E. C. (1993) Science 259, 796-798. Yang, C.-Y., Meagher, A., Huynh, B. H., Sayers, D. E., & Theil, E. C. (1987) Biochemistry 26, 497-503. Yariv, J., Kalb, A. J., Helliweg, J. R., Papiz, M. Z., Bauminger, E. R., & Norwik, I. (1988) J. Biol. Chem. 263, 13508-13510. Yu, C.-B., Harrison, P. M., Treffry, A., Quail, M. A., Arosio, P., Santambrogio, P., & Chasteen, N. D. (1995) Biochemistry 34, 7847-7853. BI970348F
